Errorless Learning and Error Correction

When to prevent errors -- and how to fix them when they happen

For paraprofessionals and the teachers who supervise them

| |

| :-: |

| The frameEvery time a student makes an error and it goes uncorrected, they practice the wrong response. Every time an error is corrected clumsily, the student may learn that errors lead to aversive interactions and begin avoiding the task. Errorless learning and systematic error correction are two tools that address this problem from opposite angles. This brief explains when each applies and how to use them. |

Why errors matter more than they seem

In everyday conversation, making a mistake and being corrected is unremarkable. In structured instruction for students with significant learning needs, errors carry more weight. When a student consistently responds incorrectly and the response is not corrected cleanly, several problems can develop:

The student rehearses the wrong answer, making it harder to learn the correct one (error strengthening)

The student learns that guessing is acceptable, reducing the quality of their responses

The task becomes associated with frustration, increasing avoidance behavior

Data reflects a mix of correct responses and prompted corrections, making it hard to read progress

The goal of both errorless learning and error correction is the same: maximize the student's experience of correct responding so that the correct response is what gets practiced and reinforced.

Errorless learning

The core idea

Errorless learning (also called errorless teaching) is an instructional strategy in which the prompt is delivered at a level that all but guarantees the student will respond correctly before they have a chance to make an error. Rather than waiting to see whether the student gets it right, the para provides a strong prompt from the outset and fades it systematically over time.

When errorless learning is indicated

When the student is learning a new skill for the first time and has no existing approximation to build on

When the student has a history of error rehearsal on similar tasks (has been practicing the wrong thing for a while)

When errors are emotionally dysregulating for the student -- some students shut down entirely after making mistakes

When the task involves a discrimination that is easily confused (e.g., similar-looking letters b/d, similar-sounding words)

When previous trial-and-error instruction has not produced progress

How to implement errorless learning

The most common approach is stimulus prompt fading or most-to-least prompt fading:

Begin with a prompt level that reliably produces the correct response (often a physical or full model prompt)

Deliver the prompt immediately -- before or simultaneously with the discriminative stimulus (SD)

Reinforce every correct response, regardless of whether a prompt was used

Systematically fade the prompt across trials or sessions according to a written fade plan

Move to the next fade level only when the student is responding accurately at the current level

Example: Teaching a student to point to their own photograph. The para begins by physically guiding the student's hand to the photo immediately when the SD is given ('Point to you'). Over time, the para fades from full physical guidance to a touch on the elbow to a gesture toward the photo to no prompt.

The tradeoff

Errorless learning can produce high accuracy quickly, but it can also create prompt dependence if the fade plan is not followed or if the student never encounters the expectation of independence. Always pair errorless learning with a written fade plan and data on independent correct responding.

Error correction

The core idea

When a student makes an error despite being given the opportunity to respond independently, the error correction procedure ensures the correct response is practiced immediately and reinforced -- while the error is not.

Common error correction procedures

The model-lead-test procedure

This three-step correction is widely used in structured literacy and ABA programs:

Model: The para provides the correct response ('That word is "there."')

Lead: The para and student produce the correct response together ('Say it with me -- there')

Test: The student is given the opportunity to respond independently ('Your turn. What word?')

After the test, move on -- do not repeat the correction loop multiple times if the student makes the same error again. Instead, flag it for the supervising teacher.

The four-step error correction (ABA)

Used in discrete trial training:

Stop: Pause the trial immediately when the error occurs

Prompt: Provide a prompt that produces the correct response

Mix/distract: Present one or two known easy trials to break the immediate error pattern

Retest: Return to the target trial and give the student another independent opportunity

Immediate model

For some programs, especially with students who are highly sensitive to extended correction sequences: the para simply models the correct response and moves on. This is less thorough than model-lead-test but produces less frustration for some students.

What NOT to do during error correction

Do not repeat the same SD multiple times without a prompt -- 'What is it? What is it? Come on, what is it?' is not an error correction procedure

Do not provide verbal praise for a prompted correct response at the same level as an independent correct response -- this creates confusion about what the reinforcement contingency is

Do not use a tone of disappointment or frustration when correcting errors

Do not skip error correction because the student seems frustrated -- the frustration is more likely to decrease when the student experiences more correct responding

Recording errors and corrections

Data on errors is as informative as data on correct responses. Common recording conventions:

\+ = independent correct response

P = prompted correct response (note the level: VP, MP, PP)

E = error (incorrect response or no response within the time limit)

If you are seeing a pattern of the same error repeating across sessions, flag it. Persistent errors that survive error correction are a signal for the program designer -- not a reason to keep running the same correction in the same way.

Scenario

The substitution error

A student learning to read decodable words consistently reads 'slip' as 'slap.' The para has been saying 'Good try, but it's "slip"' and moving on. The error continues. The supervising teacher notices the pattern and recommends switching to a full model-lead-test correction: 'That word is "slip" -- say "slip" with me -- now you say it.' The para also begins mixing in a review of the /i/ vowel sound before reading sentences containing the word. Within two weeks, the error is resolved.

| | |

| :-: | :-: |

| Try this | Watch out for |

| Match your strategy to the student: errorless for new skills or error-sensitive students; systematic correction for students working toward independence | Providing errorless prompting forever without a fade plan -- this creates learned helplessness |

| Reinforce independent correct responses more enthusiastically than prompted ones | Correcting errors by simply repeating the question louder or more urgently |

| Keep corrections calm and matter-of-fact -- the student's experience of correction should feel routine | Skipping error correction because the student seems upset -- this teaches them that errors end the demand |

| Record errors and bring patterns to the supervising teacher | Reinforcing prompted responses the same way as independent ones without the program calling for it |

| |

| :-: |

| Bottom lineError correction is not about pointing out mistakes -- it is about immediately replacing the wrong response with the right one so that the right one is what the student practices. Done well, it feels like seamless instruction, not correction. |

Page